On Mon, Aug 29, 2011 at 10:13 AM, Pieter Hintjens <p...@imatix.com> wrote: > > I'm not sure schedules are useful since we are always on a curve (more > releases when a branch is immature, fewer as it matures). But in > general we release when there've been a reasonable number of fixes and > changes. > > We've tried limited releases before, e.g. tagging on the repository, > but in fact close to no-one tests them. Experience tells us that > people only test official releases and for the most part, stable > releases only. You can see, e.g. how few issues we have on 3.0.0.
I'm of the opinion that "release candidates" are a vestige of time when releases were more expensive (e.g. burnt to CDs) and that the intent was to get everything just right. That's just not the way things work these days. I can't recall who does this, but one of the major projects uses an even-odd scheme to manage stable vs development lineages. E.g. the even releases have are more conservative and odd more relaxed. In any case, I personally don't care for the "rc" tag and am fine with rolling releases, some of which are more stable than others -- in the end, it's the list of known issues and release notes that matter. Garrett _______________________________________________ zeromq-dev mailing list zeromq-dev@lists.zeromq.org http://lists.zeromq.org/mailman/listinfo/zeromq-dev
